Praneet Bomma Email & Phone Number

London, GB

London Area, United Kingdom

Sheffield, England, United Kingdom

• Refined simulation pipeline in R by eliminating redundancies, resolving bugs, and enhancing overall performance.
• Efficiently migrated simulation library from R to Python3 and accelerated pipeline runtime by 2x with vectorization.

Greater Toronto Area, Canada

• Helped build end-to-end streaming pipeline with C++ and CUDA streams on A100 servers.
• Assisted in developing visualization tools in C++ to render real-time 3D graphics (OpenGL) overlaid on UHD video with Cinder and Unreal Engine. Visualization included rendering 3D pitch with player tracking graphics.

Greater Toronto Area, Canada

• Developed data ingestion pipeline and a deep branched architecture from scratch, integrating 3D Convolutions and LSTMs for Rain Attenuation Prediction on spatial and sensor data.
• Trained AutoEncoder and Conditional GAN for floor pattern transfer. Implemented illumination-transfer research paper and experimented with Pyvista and Blender for adding shadow effects on floor patterns.
• Enhanced object detection, segmentation, jersey classification, and pose-estimation models on 4K stream data. Developed 3D pose generation using triangulation. Initiated the development of data ingestion pipelines and.
• Orchestrated an end-to-end Realtime Football Video Analytics pipeline, efficiently processing 12 4K streams with 5 models using Deepstream, Triton and TensorRT. Leveraged Libtorch and CUDA kernels for instance.

Mumbai Area, India

• Developed row, column, and table segmentation with MaskRCNN. Implemented advanced spatial and representation rules to enhance entity extraction accuracy from invoices and receipts. Successfully improved accuracy of.
• Refined YOLOv3 and DeepSORT tracking for a Real-time Risk Monitoring Video Analytics system. Improved facemask detection with colour mode changes to training data. Optimized models with TensorRT, achieving a 2x.
• Built object detection and tracking for UAVs. Quantized models to half-precision and optimized the threading pipeline to run real-time on edge devices like Nvidia Jetson Nano and Xavier.
• Designed real-time detection and instance segmentation using Yolact++, experimented with SLAM for path planning, and crafted a novel Neural Path Planner with branched architecture incorporating human priors for.
• Developed Autonomous Navigation System for mini-trucks. Achieved 5x optimization in Occupancy Grid generated using LiDAR, increasing the speed from 4FPS to 20FPS. Integrated all components and established seamless.

Bengaluru Area, India

• Revamped the entire pipeline by implementing modular code structures, optimizing data retention practices, and establishing standardized protocols for future improvements and maintenance.
• Enhanced existing AD detection system and reduced false positives by improving pattern recognition techniques.
• Engineered an end-to-end system to expedite the process of adding new patterns and heuristic rules. Enabled instant testing on live or saved streams and facilitated rapid releases to the live system. Reduced the.
• Developed a Proof of Concept (PoC) for audio fingerprinting, aimed at enhancing the existing AD Detection system.

Mumbai, Maharashtra, India

Mumbai Area, India

Msc In Autonomous And Intelligent Systems, Information Technology, Distinction

Be - Bachelor Of Engineering, Computer Engineering, 7.23 Cgpa

Intermediate, Science